Explanation Plausibility on MDACE bigger (test)

41.4PrecisionAttInGrad

Evaluation results

Method
AttInGrad2024.06	41.4	43.2	42.2	37.7	2.7	60.3	74.8	43.7	33	44.4
AttInGrad2024.06	41.2	42.1	41.5	37	2.8	59.1	73.5	43.2	32.7	44
AttInGrad2024.06	40.9	43.9	42.2	38.1	2.5	60.8	74.2	44.1	33.2	44.7
AttInGrad2024.06	39.7	44	41.7	37.2	2.3	60.8	74.4	43.3	32.9	44.2
a∇a2024.06	38.9	36.9	37.7	32.7	16.9	52.9	63.8	40.3	30.9	41.6
Attention2024.06	38.5	37.3	37.7	33.1	22.6	52.8	61	39.1	30	40.4
a∇a2024.06	38.1	39.1	38.4	33.2	15.7	55.5	65.5	41.2	31.3	42.2
Attention2024.06	38	39.1	38.5	33.5	21.2	54.5	62.1	39.6	30.4	41
Attention2024.06	37.3	36.6	36.8	31.9	23.5	52.1	60.6	38.9	29.5	39.7
Attention2024.06	36.2	36.8	36.4	31.4	22.2	52.4	61.1	38.1	29	39
a∇a2024.06	35.7	34.4	34.9	29.5	18	50.2	61.6	37	28.5	38.3
a∇a2024.06	34.6	36.3	35.3	30	16.5	52.2	63.5	37.4	28.9	38.9
IG2024.06	34.3	33.5	33.8	27.9	5.9	49.6	64.3	35.6	28.4	38.2
InputXGrad2024.06	33.9	33	33.2	27.6	10.5	47.1	62.4	33.2	27.6	37.1
IG2024.06	33.8	33.6	33.6	27.7	5.9	50.6	65.2	35.9	28.4	38.2
Deeplift2024.06	33.7	32	32.7	26.8	10.3	46	61.8	32.2	27.2	36.5
Deeplift2024.06	33.2	32	32.5	26.3	9.4	46.1	62	31.9	27.1	36.5
IG2024.06	33	32.9	32.9	26.6	5.9	49.3	64.3	34.5	27.9	37.5
InputXGrad2024.06	32.4	31.9	32	25.7	10.1	45.6	60.8	31.6	26.8	36
InputXGrad2024.06	32.3	34.3	33.1	26.9	8.8	48.3	63.5	32.7	27.5	36.9
IG2024.06	32.2	31.8	31.8	25.7	5.7	48	63.1	34.1	26.9	36.2
InputXGrad2024.06	31.1	31.7	31.3	24.5	9.7	45.2	60.6	30.7	26.2	35.3
Deeplift2024.06	30.6	32.5	31.5	24.9	8.3	46.4	62.1	30.7	26.2	35.3
Deeplift2024.06	30.2	31.7	30.8	24	8.9	45.3	61.1	30	25.8	34.7
Rollout2024.06	1.8	25	3.3	0.4	0	25.8	32	0.4	0.5	0.7
Rollout2024.06	1.7	23.8	3.2	0.3	0	24.5	31	0.4	0.5	0.7
Rollout2024.06	1.7	24.8	3.1	0.4	0	25.4	31.3	0.4	0.5	0.7
Rollout2024.06	1.7	25.2	3.2	0.4	0	25.5	31.7	0.4	0.6	0.8
Rand2024.06	0.4	1	0.4	0.2	66.7	1.2	2.4	0.1	0.1	0.1
Rand2024.06	0.4	1	0.4	0.5	66.9	1.2	2.5	0.1	0.1	0.1
Rand2024.06	0.4	1.3	0.4	0.6	60.2	1.5	3	0.1	0.1	0.1
Rand2024.06	0.4	0.8	0.4	0.4	74	0.9	1.9	0.1	0.1	0.1